By Will Shepard, Enspiria Solutions
Global positioning systems (GPS) have revolutionized field activities for many utilities, enabling ready collection of highly accurate spatial data that can be used to support a variety of systems and objectives. However, as with anything else, there is no such thing as a “free lunch” when it comes to GPS collection.
GPS data has inherent inaccuracies and errors associated with it, the extent of which depends on the GPS receiver quality and conditions under which survey activities were taken. By understanding the quality of data which needs to be collected, a utility can make decisions about the resources to invest-both in terms of equipment costs and time spent by crews-in field collection activities, and the applications to be supported from the data collected. Ultimately, understanding accuracy and applications can optimize a utility’s use of GPS to meet its objectives.
GPS Error: Sources and Mitigation
Historically, “selective availability” was the single greatest source of GPS error. The deliberate introduction of random noise into the GPS signal by the Department of Defense to degrade accuracy for civilian (or more specifically, combatant) users while maintaining high accuracy for military users, selective availability added up to 30 meters of error to the computed receiver location. However, selective availability has been turned off since May 2000.
The next largest source of error is interference due to refraction of the signal as it passes through the upper atmosphere. Because the GPS signal is time-based, error in the amount of time for the signal to travel equates to distance error. Interference from the ionosphere can add 5 meters of error to the signal, while interference from the troposphere can add an additional half meter of error.
Multipath error occurs as the GPS signal bounces off surroundings such as buildings. When the signal bounces off walls, cars or other objects before finally arriving at the receiver, the elapsed signal travel time increases and the distance to the satellite is overestimated, introducing 0.6 meters of error. Satellite clock errors account for approximately 1.5 meters of error; although satellite clocks are highly accurate, they are still subject to some speeding or slowing, causing inaccuracy in computed signal travel time, which translates into errors in distance measurement.
Ephemeris error occurs when a satellite’s orbit drifts from the predicted orbit (due predominantly to gravitational effects). Because the predicted location of the satellite in orbit (stored in the almanac on the receiver) is used in computing receiver locations, ephemeris error can add 2.5 meters of error. Finally, the geometry of the three to six satellites (a minimum of three satellites are required to precisely pinpoint location on the Earth’s surface, and four are required for accurate elevation) being used for the measurement with respect to each other can introduce error. Fortunately, satellite geometry is completely measurable and controllable-even before hitting the field-using the GPS almanac (which contains current and projected positions of the satellites in the sky) to perform mission planning. Table 1 summarizes the sources and amounts of error from each category.
Using differential correction, the errors due to interference from the atmosphere, clock drift, ephemeris and selective availability (when and if it is turned on) from a second receiver at a highly accurate location (called a base station) can be readily determined and applied to remove these errors from the measurements. Differential correction can be applied either in real-time as the data is captured or through post-processing. Many GPS vendors offer differential correction subscription services that are integrated with their receivers to make this process painless.
Dilution of Precision (DOP) is the metric that is used to determine the amount of error due to satellite geometry. In general, the lower the DOP value, the better the geometry. Many receivers will report PDOP (positional DOP) in addition to DOP, which is DOP exclusive of clock error, but DOP is the key metric to use. The error introduced by satellite geometry is computed by multiplying the receiver precision by DOP; so, if the precision of the receiver is 10 meters, and the DOP is 3, then the best positional accuracy would be to within 30 meters. Measurements with DOP values from 1 to 3 are best for field survey activities; measurements taken with a DOP from 4 to 6 are still useable, but contain significant errors that reduce accuracy. Measurements with a DOP above 6 may meet some requirements for navigation or other purposes where relative position is more important than actual position, but for field survey purposes these measurements should be discarded. Most modern GPS receivers will allow masking to filter out data points above a threshold DOP. Because the GPS almanac contains predicted satellite locations, it can be used to perform mission planning for the optimal times to field survey-when the DOP values are at their best. Almanacs are typically only good for a day or two at most; after this, ephemeris errors affect the almanac’s quality. Thus, it’s best to download the latest almanac for mission planning purposes.
Because the GPS signal is radio-based, it is also susceptible to noise interference. Signal strength is measured through the signal-to-noise ratio (SNR); signals from satellites with an SNR of less than about 5 or 6 dB are generally considered unusable and should be discarded for field survey purposes. SNR ranges from 0 dB up to a maximum of about 35 dB; typical values range from approximately 10 to 15 dB. Like DOP, many GPS receivers will allow a mask to be applied that filters out satellites with signals below a threshold SNR. Satellite elevation is also important, and most GPS receivers will also allow a mask to be applied to filter out satellites below a minimum elevation. Typically, satellites below 15 degrees above the horizon are unreliable. This is because satellites low on the horizon must travel through more of the atmosphere and therefore suffer from more refraction, and because satellites low on the horizon may not be available to the base station for differential correction.
Multipath error is particularly difficult to deal with; because it is local to the receiver, it is not removed by differential correction. It is thus important to try to minimize multipath conditions as much as possible, but this may be difficult for utilities with facilities in urban areas. One strategy for dealing with this problem is to incorporate a laser range finder, to collect data in a nearby location with the best view of the sky and to use the laser range finder to compute the offset. In addition, other types of GNSS (Global Navigation Satellite System) such as the Russian GLONASS satellites or the European GALILEO constellation can sometimes work better than GPS for survey in areas with many buildings or trees, and many receivers today are capable of receiving signals on all three systems.
Even with differential correction, mission planning and filtering out unreliable data, it is impossible to drive error to zero with a single observation-but it can be driven practically to zero (depending on the accuracy of the receiver) with multiple observations. This is particularly important for field survey applications. When capturing data for a single location, the receiver will continuously record and apply measurements to an overall aggregate data point. Uncertainty and inaccuracy in the individual measurements “average out” over time, with the net result that the longer data is collected at a location, the more accurate the results will become. A good rule of thumb is to spend at least 5 minutes recording at each location for field surveys.
GPS Receivers and Accuracies
There are varying grades of GPS receivers in terms of accuracy; generally, these fall into consumer-grade, mapping-grade and survey-grade. As would be expected, the price point differs for each grade. Consumer-grade GPS receivers may have as much as 20 meters of error in measurement, and are typically under $500. Consumer-grade GPS receivers are typically appropriate for applications where relative position is important, such as navigation to support mobile workforce management. Mapping-grade GPS receivers range from approximately $500 to $5,000 and will have about 5 meters of error (without differential correction), while survey-grade GPS receivers will cost upward of $5,000 and will provide 1/2 meter of error (without differential correction). Consumer-grade GPS receivers typically do not support differential correction, but with differential correction mapping-grade receivers may provide meter-level accuracy, while survey-grade receivers will provide centimeter-level accuracy. Mapping-grade GPS receivers are typically adequate to support GIS requirements for (no surprise) mapping purposes, or for performing locates for readily visible structures. Survey-grade GPS receivers may be required for tasks where the highest precision is required-for example in locating underground facilities where being off by a half a meter while digging might damage other proximal infrastructure. The key to determining which type of receiver is needed is to determine the applications it will support and to determine whether the benefit gained from the accuracy justifies the cost.
GPS Applications and Precision
A common GPS goal for utilities is to be able to field collect facilities for use in the GIS. At first glance, this would appear to be a great thing-highly accurate data feeding the corporate mapping system. However, practically speaking, GPS data is often too accurate for the GIS. The underlying base map often has its own errors and inaccuracies, and laying the accurate GPS data results in facilities that are accurate with respect to the real world, but not accurate with respect to the base map. When this happens, a manhole can appear in the middle of a building. Thus it is important to consider the applications for which GPS accuracy is needed. In the case of producing system maps, it is not needed and it can be detrimental to the product (unless you’re lucky enough to have a highly accurate base map that has also been survey controlled). However, there are many other utility applications that do require the real-world location provided by GPS, with varying levels of requirements for accuracy. Table 2 summarizes common GPS applications, the level of accuracy they require, and benefits and challenges associated with each.
Another problem inherent in using GPS data to feed the GIS is whitespace management. For example, to represent a transformer and a fuse on the same pole, a utility must decide which of the three features-the transformer, the fuse or the pole-will be at the exact location. If all three are placed at their exact location in space, then the resulting map will be a blob of unreadable symbols. A typical decision is to place the structure at its correct location in space, and offset the devices. In congested areas, the challenge is even greater as facilities may be closer to each other in the field than can adequately be represented on the system map at 1:200 or 1:400 scale. In this case, the GPS accuracy must be sacrificed altogether for cartographic readability.
The most common method for striking a balance between the need for map products that are cartographically correct and readable is to place the facilities using cartographic standards, and to store the GPS coordinates using the GIS attributes. Thus for each facility, two different sets of coordinates are maintained-one in “map space” used to place the object on the map, and one for the real-world coordinates.
There are three principal applications for which high-level field survey accuracy is required: for locating buried facilities, for engineering and design and for modeling applications. For locating buried facilities, precise real-world location is critical. In the case of a facility that has been paved over, the relatively accurate GIS can get a crew to the right street or intersection, but the GPS location is required to know where to crack the pavement. Here the dual-representation of the manhole using representation on the maps gets the crews close, and the GPS coordinates stored in the attributes can be fed into a GPS unit to relocate the manhole precisely. Engineering and design is often carried out using developer-supplied survey plats, which are highly accurate. Laying out facilities using equally highly accurate locations is not only appropriate, it’s crucial. For modeling purposes, the GIS often serves as the source for populating a modeling software package. Often there is an export process from the GIS into the modeling software, and exporting the stored GPS coordinates in the export routines instead of the GIS cartographic coordinates is a simple matter of substitution in the export routines. In this way, the modeling package can take advantage of the increased accuracy to precisely model the load being carried by the system.
The key to optimal GPS field survey is to understand the errors inherent in GPS measurement and ways to mitigate those errors, and the applications for which GPS will be required-and to understand equally the applications for which GPS will not be required. With a firm understanding of accuracies and applications, a utility can optimize their use of GPS for many different applications.
Dr. Shepard is a senior consultant with Enspiria Solutions. He offers expertise in enterprise and mobile GIS and geospatially enabled applications, including work and asset management, land management, graphic work design, outage management and network analysis.